Remembrance of Things Past: Antitrust, Ideology, and the Development of Industrial Economics

50 Pages·2005·0.27 MB·English

Remembrance of Things Past: Antitrust,
Ideology, and the Development of Industrial
Economics
∗
Stephen Martin
Department of Economics
Purdue University
403 West State Street
West Lafayette, Indiana 47907-2056, USA
[email protected]
December 2005
Abstract
Key words: antitrust, industrial economics, ideology.
JEL codes: B2, L0, L4
IamgratefulforcommentsreceivedfromWilliamS.Comanor,JohnConnor,F.M.Scherer,
∗
John T. Scott, at the Università degli Studi di Lecce, and from two anonymous referees. Re-
sponsibility for errors is my own.
Contents
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2 Early Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.1 Origins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2 Monopolistic Competition . . . . . . . . . . . . . . . . . . . . . . 6
2.3 Structure-Conduct-Performance . . . . . . . . . . . . . . . . . . . 8
2.4 Did the S-C-P Paradigm Treat Market Structure as Endogenous or
Exogenous? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
3 The First Chicago School . . . . . . . . . . . . . . . . . . . . . . . . . . 12
4 The Second Chicago School . . . . . . . . . . . . . . . . . . . . . . . . 15
4.1 Chicago I and Chicago II . . . . . . . . . . . . . . . . . . . . . . . 16
4.2 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
5 Confronting the Oligopoly Problem . . . . . . . . . . . . . . . . . . . . 18
5.1 S-C-P to Chicago . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
5.2 Chicago to Game Theory . . . . . . . . . . . . . . . . . . . . . . . 26
6 Chicago Transformed . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
7 Final Thoughts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
8 Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
9 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
2
Why study the mistakes of the past?
Reaction in a faculty meeting to the suggestion that the Ph.D. pro-
gramincludearequiredcourseonthehistoryofeconomicthought.1
1. Introduction
What it is that economics can, does, or should bring to antitrust is a topic that
turnsupfairlyoftenintheliterature.2 Thequestionofwhatantitrusthasbrought
toindustrialeconomicshasreceivedlessattention, anditisthatquestiontowhich
I devote myself in this essay.
Stigler (1982, p. 6) has noted the income redistribution consequences of an-
titrust for the economics profession. Such transfers accrue to individual members
of the profession, and are not my topic. Closer to what I have in mind is the in-
valuable role of antitrust as a rich source of raw material for industrial economists
(Coase, quoted in Kitch (1983, p. 193)):
Ithinkyoucanoftenlearnmoreabouthowtheeconomicsystemworks
by reading law books and cases in law books than you can be reading
economics books because you do get descriptions of actual business
practices which are difficult to explain.
But this raw material is not without its price. I will argue that the antitrust
policy implications that can be drawn from scientific research by industrial econo-
mists have drawn ideological currents into academic debate as moths are drawn
to a flame, and that this phenomenon contributes to explaining a puzzle that has
continued for a period now going on thirty years. This puzzle is the ability of
advocates who favor a minimal role of antitrust enforcement in the economy to
portray their views, to the legal community, as generally accepted by economists,
when this claim is now and has always been manifestly incorrect.
Concerning the current views of the profession, Bolton et al. (2000, p. 2242)
write
A powerful tension has arisen between the foundations of current legal
policy and modern economic theory. The courts adhere to a static,
1I owe this heading to an anecdote of my late teacher and onetime colleague Walter Adams.
2SeeamongothersBok(1960),Sullivan(1977),Baxter(1983),Turneretal. (1983)andRowe
et al. (1984). Let me note here that anonymous referees have suggested that I include here a
discussion of the impacts of contrasting schools of thought in industrial economics on antitrust.
That is an interesting topic, and one I take up in Martin (2006).
3
non-strategic view of predatory pricing, believing this view to be an
economic consensus. This consensus, however, is one most economists
no longer accept.
U.S. courts continue to be heavilyinfluenced bywhat Posner (2001, p. 194, fn.
2) terms “orthodox ‘Chicago School’” views toward most strategic behavior, not
merely predatory pricing. Bolton et al. are correct that most economists do not
nowaccept the orthodox Chicago School analysis. Consideration of the literature
suggests that orthodox Chicago School views were never accepted by mainstream
economists.
Of course, scientific validity is not a matter of majority vote, and as John
Bates Clark wrote long ago (1887, p. 45) “Conclusions reached by valid reasoning
are always as true as the hypotheses from which they are deduced. If we admit
the fact of unlimited competition, we concede in advance many doctrines which
current opinion is now disposed to reject.” Mainstream economists have never
disputed that the policy recommendations of the orthodox Chicago School follow
as a matter of logic from the assumption that observed prices and quantities can
betreatedasgoodapproximationstolong-runcompetitiveequilibriumvalues(see
the discussion, below, of Reder (1982)). What has never been accepted by most
industrial economists is that it is appropriate to make this “good approximation”
assumption.
In Section 2, I discuss the rise of industrial economics as a branch of micro-
economics. In Section 3, I review the rise of the first Chicago School of industrial
economics, which advocated affirmative government action to obtain and main-
tain good market performance. In Section 4, I turn to the rise of the Second
Chicago School, which argued that no such government action was needed, and
thatmarketscould, withfewexceptions, betreatedasiftheywereinlong-runper-
fectly competitive equilibrium (the “good approximation” assumption). Section
5 discusses the oligopoly problem, its role in dislodging the structure-conduct-
performance paradigm by the Second Chicago School as a source of antitrust
advice and in the rise of game theoretic approaches to the analysis of imperfectly
competitive markets by economists. Section 6 discusses the evolution of the Sec-
ond Chicago School in the face of the evident failure of mainstream economists
to accept the good approximation assumption. Section 7 concludes with a few
remarks on ideology and its impact on industrial economics.
4
2. Early Development of Industrial Economics3
2.1. Origins
Industrial economics is generally said to have emerged as a distinct branch of
microeconomicswithEdwardS. Mason’sHarvardseminarof the1930s(Markham
and Papanek, 1970, pp. vii—viii), but the topics that occupy industrial economists
have concerned economists since before the emergence of political economy as a
distinct branch of the social sciences in 1776.
Policy questions that remain at the heart of industrial organization were the
subject of widespread academic and popular debate in the United States between
1880and1900, adebatethatcontinuedatonlyaslightlylessintenselevelbetween
1900 and 1920. Seven of the first 10 presidents of the American Economic As-
sociation played active roles in this debate.4 Marshall’s 1919 Industry and Trade
made international comparisons in industrial organization and drew conclusions
foreconomicdevelopment. Likethebourgeoisgentlemanwhospokeprosewithout
realizing it, economists who studied “railway problems” (Ripley, 1907) or “trust
and corporation problems” (Burns, 1937, p. 663) studied industrial economics in
everything but name.
But they did so using analytical tools that seemed to them to be ill-suited to
the task. The mainstream price theory of the early twentieth century consisted of
a theory of competitive markets and a theory of monopoly, with a vast wasteland
in between.5 This theory of competitive markets was not the modern model
of perfect competition, but its Marshallian predecessor. To classify a market
as competitive in this sense required only that it “would be possible for other
businesses to produce a commodity with the same technical specifications as the
product of any particular firm, and to offer it for sale to that firm’s customers.”
If this condition were met (Andrews, 1951, pp. 141—142), “the possibility of entry
of other producers would ensure that long-run price would be equal to the normal
average cost of production.”
3For discussions of the development of the field of industrial economics, see Bain (1949a, pp.
129—133), E. T. Grether (1970), Phillips and Stevenson (1974), David Dale Martin (1976), Hay
and Morris (1979), Schmalensee (1982, 1987, 1988), Davies and Lyons (1989), Bonanno and
Brandolini (1990), and, from a perspective of legal scholarship, Hovenkamp (1991, Chapter 22).
4TheseareFrancisA.Walker,JohnBatesClark,HenryC.Adams,ArthurT.Hadley,Richard
T. Ely, Edwin R. A. Seligman, and Jeremiah W. Jenks. The works of two others (Tassig and
Patten)ontariffpolicytouchedupontopicsthatwouldnowbeclassifiedinindustrialeconomics.
5Schumpeter (1934, p. 249); Bain (1944, p. 4); Stigler (1949, p. 12); Andrews (1951, p. 141;
1952, p. 72); Schneider (1967, p. 139).
5
Economists of the period were aware of the disconnect between the implica-
tions of this theory of competitive markets and the industrial world around them
(Marshall, 1925 [written in 1890], p. 268):
it is chiefly from America that a cry has been coming with constantly
increasing force for the last fifteen years or more, that in manufactures
free competition favours the growth of large firms ...; that such firms,
if driven into a corner, will bid for customat any sacrifice; that, rather
thannotselltheirgoodsatall,theywillsellthemat[marginalcost]...,
which is sometimes very little; that, when there is not enough work
for all, these manufacturers will turn their bidding recklessly against
one another, and will lower prices so far that the weaker of them will
be killed out, and all of them injured; so that when trade revives they
willbeable,evenwithoutanycombinationamongstthemselves,toput
up prices to a high level; that these intense fluctuations injure both
the public and the producers; and the producers, being themselves
comparatively few in number, are irresistibly drawn to some of those
many kinds of combinations to which, nowadays, the name Trust is
commonly ... applied.
Spurred by this perceived disconnect, some economists developed new theoretical
tools, while others turned to empirical approaches.
2.2. Monopolistic Competition
An initial theoretical response was the attempt to refine the cost curve apparatus
of Marshall’s theories of the firm and of industry supply. Sraffa, commenting on
these efforts, drew attention to the importance of the demand side of the market
for market performance (1926, p. 543):
The chief obstacle against which [businessmen] have to contend when
they want gradually to increase their production does not lie in the
cost of production–which, indeed, generally favours them in that
direction–but in the difficulty of selling the larger quantity of goods
without reducing the price, or without having to face increased mar-
keting expenses. This necessity of reducing prices in order to sell a
larger quantity of one’s own product is only an aspect of the usual
descending demand curve, with the difference that instead of concern-
ing the whole of a commodity, whatever its origin, it relates only to
6
the goods produced by a particular firm; and the marketing expenses
necessaryfor the extensionof its market are merelycostlyefforts ...to
increase the willingness of the market to buy from it–that is, to raise
that demand curve artificially.
Two books published in 1933, Edward Chamberlin’s The Theory of Monopo-
listic Competition and Joan Robinson’s The Economics of Imperfect Competition,
followed up on Sraffa’s theme.6 Both put forward analytical frameworks that in-
corporated firm-specific downward sloping demand curves and costly sales efforts
as essential aspects of the firm’s environment.
Robinson (1969, p. xiii) acknowledges her debt to Sraffa (1926) and the lit-
erature of which it was a part. Chamberlin (1961, pp. 517—518) denies any
connection between his work and the cost curve controversy. He traces it instead
toamuchearlierdebateonrailroadrates, adebatethatdrewhisattentiontoboth
oligopoly and product differentiation as features of industrial markets.7 Railroads
were the first wave in the rise of large-scale industry in the United States, and
to this extent Chamberlin’s work may be more directly connected than that of
Robinson with the underlying industrial developments that prompted economists
to explore new analytical frameworks.
Samuelson (1967, p. 138) writes of the theory of monopolistic competition as
a revolution, leading “economists into a new land,” the land of imperfect compe-
tition (Samuelson 1967b, p. 108, fn. 5):
If the real world displays the variety of behavior that the Cham-
berlin-Robinson models permit ...then reality will falsify many of
6On the relation between the two works, see White (1936) and Fisher (1989, p. 114, fn. 2,
emphasis in original):
It is interesting to note that Joan Robinson, in her The Economics of Imperfect
Competition (1933) which is often paired with Chamberlin’s book, simply failed
to understand the oligopoly problem altogether. She assumed (p. 21) that the
behavior of each oligopolist can be modelled by creating a demand curve taking
the optimal reactions of rivals into account and then having the oligopolist set
marginal revenue equal to marginal cost. This totally begs the question of what
thoseoptimalreactionsare–andthefactthatonecannotknowtheanswertothat
beforecreatingthetheoryisthecentralcoreoftheoligopolyproblem. Chamberlin
made considerable efforts to differentiate his product from that of Joan Robinson.
In the oligopoly dimension, at least, he was right.
7The relation between the railroad rate controversy and Chamberlin’s work is discussed by
Ekelund and Hébert (1990).
7
the important qualitative and quantitative predictions of the compet-
itive model. Hence, by the pragmatic test of prediction adequacy, the
perfect-competition model fails to be an adequate approximation.
Stigler offers a more measured assessment (1949, p. 24):
The general contribution of the theory of monopolistic competition
...seems to me indisputable: it has led to reorientation and refine-
ment of our thinking on monopoly. We are now more careful to pay
attention to the logical niceties of definitions of industries and com-
modities. Wearenowmorecarefultoapplymonopolytheorywhereit
is appropriate. The importance of the trade mark and of advertising,
and the need for the study of product structure and evolution, have
become more generally recognized.
Most economists would probably credit the theory of monopolistic competi-
tion with a greater impact than Stigler is willing to concede. If it appears less
than revolutionary to contemporary economists, that may well be because they
have grown up accustomed to having monopolistic competition as part of their
intellectual landscape.8
2.3. Structure-Conduct-Performance
The analytical framework that came out of the 1930s was the structure-conduct-
performance (S-C-P) paradigm. It was formulated in literary rather than math-
ematical form, and it held center stage in industrial economics for some 40 years.
TheeconomistswhoerectedtheS-C-Pframeworkwereinterestedinexplaining
the wayprices were determined in imperfectlycompetitive markets. This interest
was explicitly motivated by contemporary industrial developments (Mason, 1939,
p. 63):
The growth of corporate bureaucracies (with the consequent institu-
tionalization of management decisions), the separation of ownership
8Forelementsofonedebateonmonopolisticcompetition,seeArchibald(1961,1963),Stigler
(1963), Friedman (1963), and also Chamberlin (1957, Chapter 15). I do not pursue this and
other such skermishes, as they proved to be false starts in the debate between Chicago and the
rest of the profession. The seed planted by Chamberlin did not fully flower until much later
(Dixit and Stiglitz, 1977; Salop, 1979; Wolinsky, 1986).
8
from control, and the growing influence of labor organization on pol-
icy making are all factors “internal to the firm” which may and do
affect its reaction to market situations.
Their view was that the models of competitive markets and of monopoly that
economists had to work with were not suited for this purpose (Mason, 1939, p.
61):
Inperfectmarkets,whethermonopolisticorcompetitive,priceishardly
a matter of judgment and where there is no judgment there is no pol-
icy. The area of price policy, then, embraces the deliberative action
of buyers and sellers able to influence price; that is to say, it covers
practically the whole field of industrial prices.
They rejected early formal theoretical models of imperfectly competitive mar-
kets as inapplicable in practice (Mason, 1939, p. 62):9
It would no doubt be extremely convenient if economists knew the
shape of individual demand and cost curves and could proceed forth-
with,bycomparisonsofpriceandmarginalcost, toconclusionsregard-
ing the existing degree of monopoly power. The extent to which the
monopoly theorists, however, refrain from an empirical application of
their formulae is rather striking. The alternative, if more pedestrian,
route follows the direction of ascertainable facts and makes use only
of empirically applicable concepts.
Theyalsorejectedthethen-commoninstitutionalapproachtoindustrystudies,
which they felt was primarily descriptive (Burns, 1937, p. 664, emphasis added):
studiesofparticularindustriesassumedaconventionalpattern. ...The
technical processes of production were described. The organization of
the industry was discussed in terms of the size and location of plants,
the scope of ownership control (the size and extent of integration of
firms), theorganizationofmarketing,laborconditions, andthehistory
of mergers in the industry. ...The discussion of wages and possibly
profits implied an interest in the functioning of the industry, but the
aspect of its functioning most vital to theorists and purchasers, namely
its price policy, received scant attention.
9See similarly Bain (1944, p. 5). For later calls for “a return to the data,” see Andrews
(1951, p. 172, 1952, p. 75) and Coase (1972).
9
Mason (1939, p. 61) specifically rejects the institutionalist approach.
Economists of the time called for a general analytical framework (Burns, 1937,
p. 665):10
The primary necessity ...is some broad framework within which price
behavior can be analyzed in various industries. It must explain the
relationship between the organization of production and distribution
and the behavior of buyers and of prices.
The structure-conduct-performance paradigm developed out of Mason’s semi-
nar. It was the organizing framework of research in industrial organization from
the 1930s to the 1970s, and was the basis of the two successive leading textbooks
in the field.11
The earliest research in what became the S-C-P tradition were book-length
studies of single industries. Subsequently, Bain (1956) published a comparative
study of a small number of industries. He did not employ regression analysis.
Other industrial economists soon turned to the econometric analysis of cross-
section samples of industry data, first of small numbers of industries, and later of
large samples covering essentially all manufacturing.
2.4. Did the S-C-P Paradigm Treat Market Structure as Endogenous
or Exogenous?
Because of the role the issue played in the dislodging of the S-C-P approach by
what I will call the Second Chicago School and the subsequent dislodging of the
Second Chicago School by game-theoretic analysis, I wish to deal explicitly with
thequestionwhethertheS-C-Pparadigmtreatedmarketstructureasbeingdeter-
mined by economic forces, or as exogenous, determined outside the marketplace?
Some economists have taken the view that the S-C-P approach paid scant
attention to the determinants of market structure. McGee (1988, p. 2, emphasis
in original) takes the view that causation in the S-C-P paradigm was mostly
in one direction: “In the beginning, most economists seem to have believed that
the structure-conduct-performance relationshipwaslargelyoraltogether oneway:
to a significant degree the structure of an industry determines the conduct of
firms in it; and how firms behave to a significant degree determines how well the
industry performs.” Davies and Lyons (1996, p. 89) write that economists who
10See similarly Mason (1939, p. 61) and Andrews (1952, p. 75).
11Bain (1959); Scherer (1970, 1980), and Scherer and Ross (1990).
10